Peritoneal Dialysis After Cardiothoracic Surgery: Do It!

Cardiovascular disease remains the most common comorbid condition and cause of mortality in patients with chronic kidney disease (1). As a consequence, increasingly large numbers of end-stage renal disease (ESRD) patients are undergoing coronary artery revascularization procedures. In this patient population, coronary artery bypass grafting (CABG) appears to be preferred over percutaneous coronary artery intervention. In a cohort of patients with ESRD [including both hemodialysis (HD) and peritoneal dialysis (PD) patients], a much lower 3-year mortality risk was observed with CABG (hazard ratio: 0.39; p = 0.0006) than with percutaneous artery intervention (2). In patients with chronic kidney disease stages 3 – 5 (ESRD patients were not analyzed separately) who had severe (3-vessel) coronary artery disease, that difference held true even when drug-eluting stents were used (3). As a result, CABG is more frequently being performed in patients with ESRD.

Use of PD after cardiothoracic surgery (CTS) offers several potential advantages over HD. Those advantages include the avoidance of hemodynamic instability, bacteremia, and bleeding secondary to the use of heparin [although heparin use can often be circumvented, particularly in the setting of the intensive care unit (ICU), with a dialysis nurse dedicated specifically to performing the HD procedure]. In addition, the equipment required for PD is significantly less complex than that required for HD, such that, in many institutions (although certainly not all), the performance of PD does not demand additional personnel beyond the ICU nurse: automated PD is initiated by the PD nurse and is then monitored by the ICU nurse, with the PD nurse available by telephone to address technical concerns should they arise. It is somewhat surprising, therefore, that the utilization of PD after CABG or other cardiothoracic procedures has not been extensively studied.

Several reports on outcomes in patients with ESRD undergoing CTS have been published (4–6). However, in each one, the total number of patients was small, and the number of PD patients was extremely modest. None of the studies directly compared outcomes between HD and PD patients. The largest previous report in which outcomes for patients managed with PD after CTS were compared with outcomes in patients managed with HD (65 on HD, 40 on PD) originated from Toronto, and all of the study patients underwent CABG only (7). Preoperative demographic characteristics of the HD and PD patient cohorts were not provided, and one cannot therefore determine whether the groups were comparable. However, in that study, PD was associated with poor outcomes. Peritonitis developed in 12.5% of the PD patients, and a peritoneal–pleural leak was detected in 10%. Of greatest concern was the dramatically increased risk of in-hospital mortality for the patients undergoing PD, with an adjusted odds ratio of 22.58 (p = 0.015). The reason or reasons for that increase in mortality could not be ascertained. Importantly, 10 of the 40 PD patients were switched to either HD or continuous venovenous HD during the postoperative period, and of the 7 in-hospital deaths among the PD patients, 6 occurred in this latter group who were switched to HD. Thus, it is not clear that the increased risk of mortality in the PD patients is properly attributable to the PD modality.

In this issue of Peritoneal Dialysis International, Kumar et al. report a case–control series of 36 PD patients who underwent CTS over a period of 15 years (8). Each PD patient was paired with 2 HD patients who underwent CTS during the same period. The controls were chosen based on concordance with the PD patients for age, presence or absence of diabetes, and modified Charleston comorbidity index. In addition, the patients and controls proved to be well-matched for sex and race. The likelihood of the surgery being done electively rather than urgently was slightly higher in the HD patients, which might have favored superior outcomes in that group. However, that advantage was likely offset by dialysis vintage, which, in the range reported in the study, is recognized as an important indicator of mortality (9), and which was slightly shorter in the PD patients, potentially favoring outcomes in that group. However, neither of those trends approached statistical significance (p > 0.35 for both). In addition, unlike the Toronto report, which was restricted to patients undergoing CABG alone (7), the present series included patients undergoing CABG alone, valve replacement alone, or a combination (a distribution that also proved to be similar in the patient and control groups).

For all parameters assessed, outcomes in the PD patients compared very favorably with those in the HD patients; in some instances, the outcomes were even better. The primary study outcome, operative mortality at 30 days, was identical in the PD and HD patients at 10% and 11% respectively, and it was within the range reported in previous studies of ESRD patients undergoing CTS (10,11). Similarly, 2-year survival was at least as good in the PD patients as in the HD patients at 68.6% and 65.7% respectively. The differences were not statistically significant. Median length of stay in the cardiac surgical unit was significantly shorter for PD patients (2 days vs 4 days, p = 0.05), although overall length of stay in the hospital did not differ. Use of PD was associated with strong trends approaching statistical significance for shorter median intubation time (p = 0.06) and fewer postoperative infections (p = 0.08). Specifically, in the PD group, there appeared to be fewer episodes of pneumonia (1 in 36 patients vs 6 in 72 patients—perhaps related to shorter median intubation time) and fewer episodes of bacteremia (0 in 36 patients vs 3 in 72 patients). Unfortunately, the nature of the HD vascular access—central venous catheter or arteriovenous fistula or graft—in the 3 HD patients who developed bacteremia is not provided. In contrast to the Toronto study, the present study reported that no PD patient developed peritonitis during the observation period. The cumulative incidence of any postoperative complication was significantly lower in the PD patients than in the HD controls (28% vs 50%, p = 0.046).

Our own much more modest experience accords with that reported by Kumar et al. Over the past 3 years, 4 of our PD patients underwent CTS, and all were managed successfully in the postoperative period with automated PD. No patient developed peritonitis or a dialysate leak, no patient required transition to HD or continuous renal replacement therapy, and the 30-day survival was 100%. Of those 4 patients, 2 underwent CABG alone, 1 underwent aortic valve replacement alone, and 1 underwent aortic valve replacement with replacement of the aortic root. Thus, even patients undergoing extremely complex CTS procedures can be managed postoperatively with PD.

The present study does, however, have a few limitations. First, this is a single-center study from an institution that functions as the referral center for a large geographic area. It is therefore unclear whether the results obtained may be extrapolated to other—especially smaller—centers. Of course, as is true of many medical procedures in general, and of PD in particular, larger patient volumes are often associated with superior outcomes (12,13), which suggests that the model used at the study institution should be adopted in other geographic areas as well. That is, PD patients (and probably those on HD too) who require major CTS procedures should be funneled to a regional referral center (which will then gain substantial experience with such patients), rather than have each of several smaller centers operate on small numbers of these patients. It is likely that patients receiving their surgery at these larger referral centers will enjoy superior outcomes.

The authors report that 2 PD patients required conversion to HD during the postoperative period (and that only 1 HD patient required conversion to continuous renal replacement therapy). Of the conversions to HD, one was necessitated by dialysate leakage. Unfortunately, we are told neither the location of the leak, nor whether it was related to the surgical procedure or was independent thereof. When considering the use of PD after CTS, it is important to have a preoperative discussion with the surgeon and to ensure that, if at all possible, the surgical incision itself and all drains, chest tubes, and so on avoid penetrating the diaphragm. Failure to maintain the integrity of the diaphragm is indeed very likely to result in dialysate leakage and thereby preclude the use of PD in the postoperative period. The second conversion was said to be for “uncontrolled azotemia,” although the exact nature of the clinical concerns was not defined. It should be noted that, although all PD patients were treated with automated PD, a detailed description of the regimens utilized is not provided; the only comment is that patients “were generally continued on their existing dialysis prescription.” More information regarding the number of cycles, dwell volumes, glucose concentrations, and use (or not) of icodextrin would have been desirable.

It is unsurprising that standard PD might not suffice for all patients in the immediate postoperative period after a major cardiothoracic procedure. It would have been of great interest to see if the use of high-volume PD with more numerous and frequent exchanges targeting a daily Kt/V of 0.65 might have obviated the need for conversion to HD. The high-volume technique has been used successfully, with outcomes comparable to those obtained with 6-times-weekly intermittent HD, in patients with acute kidney injury, including those with sepsis and extreme catabolic states (14).

Finally, although the authors note that 6 of the PD patients were treated with diuretics during their hospital stay, comparable data are not provided for the HD patients. Similarly, information regarding average daily fluid removal—by dialysis and by urine output—is lacking for both groups.

Notwithstanding the above concerns (which in my opinion do not substantially detract from the validity of the report), the authors of the present study are to be congratulated for their important contribution to this field. They have convincingly demonstrated that PD patients may undergo even the most complex CTS procedures and then safely be managed postoperatively using PD alone. In the past, nephrologists have frequently needed to “do battle” with cardiothoracic surgeons who were loathe to embrace this course of management. Being now armed with this important and well-performed study, nephrologists should be able to care for patients who require major CTS procedures without obligating those patients to undergo HD, with its attendant risks of hemodynamic instability and, for those without permanent vascular access, catheter-related bacteremia.

DISCLOSURES

The author has no financial conflicts of interest relevant to the present work.